Edit one line, pay for one line
Regeneration charges only what you changed. Unchanged lines replay from cache — free and instant.
What changes when the studio runs the read
Four jobs the home page teases — directed here with the mechanism, the limit, and the doubt each one raises.
How it works
From script to finished audio in three steps
Bring the words
Paste a script, generate one with AI, or transcribe a recording.
Cast & direct
Assign a voice per speaker, add performance tags, polish with AI assist.
Export & publish
One seamless track, loudness presets, captions — download when ready.
Two speakers. One track.
Write labels like a script, assign a catalog voice to each role, and export one level-matched file with a beat at every turn.
Label characters are not charged. Unlabeled paragraphs use your default voice.
Is this a cloner? Can I upload a real person’s voice?
Catalog voices only. We do not clone real people.
Script
VESPER
Depth check — we just passed four thousand meters.
CRANE
Steady as she goes. Kill the floodlights for a moment.
The line you would regenerate
VESPER
Whoa. The whole trench wall is glowing.
CRANE
Bioluminescence. Thousands of them, all around us.
VESPER
Nobody has ever filmed this species alive.
CRANE
Then let's make sure the cameras are rolling.
Change one line. Keep the rest.
Regeneration charges only the paragraph you edited. Unchanged lines replay from cache — no second bill for the chapter you already finished.
Identical text, voice, and speed is a cache hit. If a paid job fails, characters go back automatically.
Will a long chapter drift, or will I pay twice for one typo?
We split on sentence boundaries and stitch one file. You are not billed again for lines you did not change.
Script
Para 1
Every city has a building people walk past without ever looking up. In ours, it is the old telephone exchange on Mercer Street — nine floors of red brick, and for sixty years, the place where every conversation in the city passed through.
Para 2
Inside, four hundred operators worked the switchboards around the clock. They knew which doctor to wake at two in the morning. They knew which shop still had bread. When the river flooded in fifty-three, it was an operator named Ruth who called every house on the low road before the water reached them.
Para 3
The exchange closed in 1987. The switchboards went to a museum, the building went quiet, and the city kept talking to itself through cables nobody thinks about, in a place nobody looks up at.
Only this paragraph is charged again
Recover the words, recast the voices.
Upload audio or video, clean the transcript, translate if you need, then speak it with catalog voices — without rebuilding the script by hand.
You leave with audio and captions. This is not video dubbing and not a lip-sync tool.
Will this dub my video and keep the picture in sync?
No. We do not mux video. You get a new voice-over and an SRT.
Upload the recording
Audio or video. We pull a transcript you can edit — including speaker labels when diarization is available.
Clean the words
Fix names and cuts in the transcript. Optional translate for spoken delivery, not a subtitle dump.
Cast and export
Assign catalog voices and download the new audio plus captions. We do not mux video.
Finish a chapter, not a clip.
Paste a long script — split on sentences, generated in parallel, stitched to one file — or upload a CSV (one row per chapter or episode) and download a zip. A bad row fails alone.
Captions ship next to the audio, on every plan. Loudness presets: video -14 LUFS, podcast -16 LUFS.
Will a chapter-length script come back as one file, or a pile of clips?
One file. The editor is built for that length; the demo on the home page is the short version of the same pipeline.
CSV
filename,text,voice_id chapter-01.mp3,The marsh road was empty before dawn., chapter-02.mp3,She kept the radio low so the house would stay asleep.,orion
Start from a topic when the script is not ready.
Topic, length, tone, and language in — a spoken narration script out, ready to send to the editor.
This is how some people bring the words. It is not the reason to pick the studio over a cloner.
Does the writer invent facts I will have to strip out?
Treat it as a draft. Edit it in the same project before you generate speech.
Why creators switch
Built for real production work
Long scripts, one seamless file
Paste a whole chapter: sentences split, generate in parallel, stitch clean, with loudness presets for video (−14 LUFS) and podcast (−16 LUFS).
Captions included
An .srt with correct timing next to every voice-over, on every plan — including free.
Names spoken your way
Character names, invented places, jargon: define the pronunciation once and every voice in the cast says it right.
Direct the performance
Insert laughs, pauses, emphasis and more — or let AI Director suggest speech tags and delivery packs you can accept or edit.
Failures refund automatically
If a generation fails after credits were charged, the characters go straight back to your balance — no support ticket.
Hear the read before you re-record anything.
A free account opens the editor, captions, and the rest of the monthly allowance.
Free to start
Then from $7/month for about 3 hours of audio. One-time top-up packs never expire.